Weekly digest · new literature
New papers in environmental science
A weekly sweep of a fixed list of environmental journals. Each card carries a short summary and links to the paper's DOI; the rule that decided which papers are here is printed above the list, exactly as the build recorded it. The summaries are written by a language model and can be wrong; the linked paper is the authoritative version. Cards marked abstract excerpt carry the publisher's own abstract text instead of a machine summary, and cards marked from the title alone flag a summary the model wrote with no abstract to work from, because the publisher deposited none.
Loading the digest…
Topic
Journal
Order
Loading the digest…
What this page is, and what it is not
Where it comes from
A scheduled job queries a fixed list of journals once a week and writes a single JSON file that this page renders. The journals polled for this build are listed at the bottom of the page, each linked to its own site, with how much each one contributed.
What decided which papers are on the page is stated above the list, in the build's own words, and that sentence is the whole of it. This page adds nothing to it: no importance score, no recommendation, no judgement of quality. The default order is the day each paper was listed by Crossref or OpenAlex, newest first, and the alternative is alphabetical by journal.
The open access marker appears only where the record says so outright. Its absence means the record does not assert open access, which is not the same as the paper being paywalled for you; institutional and repository copies are invisible here.
About the summaries
Cards marked AI summary were written by a language model from the paper's own abstract. Models paraphrase, and paraphrase introduces error: an effect size can move, a hedge can vanish, a conditional finding can come out flat. Read the summary as a pointer to the paper, never as a substitute for reading it.
Cards marked AI summary · from the title alone are weaker still. Several large publishers, ACS and Elsevier among them, do not deposit abstracts to Crossref, so for those papers the model was handed the title and nothing else. What it produced is an expansion of the title, not a reading of the paper, and it is labelled that way so the three kinds of card are distinguishable at a glance.
Cards marked abstract excerpt carry the publisher's own abstract text, copied unchanged. The three are never mixed and never relabelled as each other. Titles are always the publisher's own, never rewritten.
Author lists are truncated to the first three names, with the true total shown after them. The paragraph at the top of the page, when one is present, was written by the same language model, from the papers below and nothing else, and every claim in it should be traceable to one of them.